22 Oct
2007
22 Oct
'07
6:23 p.m.
"Yitzchak Gale"
Henning Thielemann wrote:
HXT uses Parsec, which is strict. I had a look at using HXT awhile ago. Parsec is the least of the problems. HXT stores the XML as an explicit tree in memory, where the head has explict references to the children. This means that the whole XML tree is stored in memory until the last child is processed. Also this tree is stored ineffeciently. Everything as non shared Haskell strings. My experience is that a 30MB file (which is quite small for an XML file) can NOT be processed with 2GB memory.
Rene.